FERROELECTRICS ‐ APPLICATIONS Edited by Mickaël Lallart
Ferroelectrics - Applications Edited by Mickaël Lallart
Published by InTech Janeza Trdine 9, 51000 Rijeka, Croatia Copyright © 2011 InTech All chapters are Open Access articles distributed under the Creative Commons Non Commercial Share Alike Attribution 3.0 license, which permits to copy, distribute, transmit, and adapt the work in any medium, so long as the original work is properly cited. After this work has been published by InTech, authors have the right to republish it, in whole or part, in any publication of which they are the author, and to make other personal use of the work. Any republication, referencing or personal use of the work must explicitly identify the original source. Statements and opinions expressed in the chapters are these of the individual contributors and not necessarily those of the editors or publisher. No responsibility is accepted for the accuracy of information contained in the published articles. The publisher assumes no responsibility for any damage or injury to persons or property arising out of the use of any materials, instructions, methods or ideas contained in the book. Publishing Process Manager Silvia Vlase Technical Editor Teodora Smiljanic Cover Designer Jan Hyrat Image Copyright Noel Powell, Schaumburg, 2010. Used under license from Shutterstock.com First published June, 2011 Printed in Croatia A free online edition of this book is available at www.intechopen.com Additional hard copies can be obtained from
[email protected] Ferroelectrics - Applications, Edited by Mickaël Lallart p. cm. ISBN 978-953-307-456-6
free online editions of InTech Books and Journals can be found at www.intechopen.com
Contents Preface IX Part 1
Sensors and Actuators
1
Chapter 1
Giant k31 Relaxor Single-Crystal Plate and Their Applications 3 Toshio Ogawa
Chapter 2
MEMS Based on Thin Ferroelectric Layers 35 Igor L. Baginsky and Edward G. Kostsov
Chapter 3
Periodically Poled Acoustic Wave-Guide and Transducers for Radio-Frequency Applications 59 Sylvain Ballandras, Emilie Courjon, Florent Bassignot, Gwenn Ulliac, Jérôme Hauden, Julien Garcia, Thierry Laroche and William Daniau
Chapter 4
Ferroelectric Polymer for Bio-Sonar Replica 75 Antonino S. Fiorillo and Salvatore A. Pullano
Chapter 5
Ferroelectric Materials for Small-Scale Energy Harvesting Devices and Green Energy Products Mickaël Lallart and Daniel Guyomar
Part 2
Memories
95
115
Chapter 6
Future Memory Technology and Ferroelectric Memory as an Ultimate Memory Solution 117 Kinam Kim and Dong Jin Jung
Chapter 7
Ultrahigh Density Probe-based Storage Using Ferroelectric Thin Films 157 Noureddine Tayebi and Yuegang Zhang
VI
Contents
Chapter 8
Fabrication and Study on One-Transistor-Capacitor Structure of Nonvolatile Random Access Memory TFT Devices Using Ferroelectric Gated Oxide Film 179 Chien-Min Cheng, Kai-Huang Chen, Chun-Cheng Lin, Ying-Chung Chen, Chih-Sheng Chen and Ping-Kuan Chang
Chapter 9
Ferroelectric Copolymer-Based Plastic Memory Transistos 195 Sung-Min Yoon, Shinhyuk Yang, Soon-Won Jung, Sang-Hee Ko Park, Chun-Won Byun, Min-Ki Ryu, Himchan Oh, Chi-Sun Hwang, Kyoung-Ik Cho and Byoung-Gon Yu
Chapter 10
Use of FRAM Memories in Spacecrafts 213 Claudio Sansoè and Maurizio Tranchero
Chapter 11
Adaptive Boolean Logic Using Ferroelectrics Capacitors as Basic Units of Artificial Neurons 231 Alan P. O. da Silva, Cicília R. M. Leite, Ana M. G. Guerreiro, Carlos A. Paz de Araujo and Larry McMillan
Preface Ferroelectricity has been one of the most used and studied phenomena in both scien‐ tific and industrial communities. Properties of ferroelectrics materials make them par‐ ticularly suitable for a wide range of applications, ranging from sensors and actuators to optical or memory devices. Since the discovery of ferroelectricity in Rochelle Salt (which used to be used since 1665) in 1921 by J. Valasek, numerous applications using such an effect have been developed. First employed in large majority in sonars in the middle of the 20th century, ferroelectric materials have been able to be adapted to more and more systems in our daily life (ultrasound or thermal imaging, accelerometers, gy‐ roscopes, filters…), and promising breakthrough applications are still under develop‐ ment (non‐volatile memory, optical devices…), making ferroelectrics one of tomor‐ row’s most important materials. The purpose of this collection is to present an up‐to‐date view of ferroelectricity and its applications, and is divided into four books:
Material Aspects, describing ways to select and process materials to make them ferroelectric.
Physical Effects, aiming at explaining the underlying mechanisms in ferroelec‐ tric materials and effects that arise from their particular properties.
Characterization and Modeling, giving an overview of how to quantify the mechanisms of ferroelectric materials (both in microscopic and macroscopic approaches) and to predict their performance.
Applications, showing breakthrough use of ferroelectrics.
Authors of each chapter have been selected according to their scientific work and their contributions to the community, ensuring high‐quality contents. The present volume focuses on the applications of ferroelectric materials, describing innovative systems that use ferroelectricity. The current use of such devices as sensors and actuators, in the field of acoustics, MEMS, micromotors and energy harvesting will be presented in chapters 1 to 5. The next section proposes a particular emphasis
X
Preface
on the application of ferroelectric materials as transistors and memory devices (chap‐ ters 6 to 11), showing one of the future breakthrough uses of these materials. I sincerely hope you will find this book as enjoyable to read as it was to edit, and that it will help your research and/or give new ideas in the wide field of ferroelectric mate‐ rials. Finally, I would like to take the opportunity of writing this preface to thank all the au‐ thors for their high quality contributions, as well as the InTech publishing team (and especially the publishing process manager Ms. Silvia Vlase) for their outstanding sup‐ port. June 2011 Dr. Mickaël Lallart INSA Lyon, Villeurbanne, France
Part 1 Sensors and Actuators
1 Giant k31 Relaxor Single-Crystal Plate and Their Applications Toshio Ogawa
Department of Electrical and Electronic Engineering, Shizuoka Institute of Science and Technology, Japan 1. Introduction
Typical ferroelectric ceramics, lead zirconate titanate (PZT) ceramics are widely used for devices of electrical-mechanical energy conversion devices such as sensors and actuators, which correspond to the five senses and foot & hand of human being. Recently, these devices spread out in the computer controlled fields, for example, robotics and mechatronics. The research and development of ferroelectric ceramics, particularly PZT ceramics, have mainly focused on the material compositions to realize new electronic devices utilizing their piezoelectric properties. Many researchers in companies and institutes have carried out R & D on such chemical compositions since the discovery of piezoelectricity in PZT ceramics by Jaffe et al. in 1954. On the other hand, through the new research on DC poling field dependence of ferroelectric properties in PZT ceramics, the poling field has become an effective tool for evaluation and control of the domain structures, which fix the dielectric and ferroelectric properties of PZT ceramics. Therefore, PZT ceramics with different domain structures can be fabricated even though the ceramic compositions remain the same. These ceramics are called poling field domain controlled ceramic. It is thought that the domain controlled ceramics will lead to a breakthrough and the appearance of new ferroelectric properties. The study on the clarification of relationships between [compositions] vs [poling fields] vs [dielectric and piezoelectric properties] in hard and soft PZT ceramics was applied to other ferroelectric materials of lead titanate ceramics, lead-free ceramics such as barium titanate, alkali bismuth niobate, alkali bismuth titanate ceramics and relaxor single crystals of Pb[(Zn1/3Nb2/3)0.91Ti0.09]O3 (PZNT91/09) and Pb[(Mg1/3Nb2/3)0.74Ti0.26]O3 (PMNT74/26) compositions. This chapter describes how can be achieved the new ferroelectric properties such as giant transverse-mode electromecanical coupling factor of k31 over 80% and piezoelectric strain d31 constant of -2000 pC/N in PZNT91/09 and PMNT74/26 single crystals realized a monodomain single crystal by accurately controlling the domain structures. In addition, highefficiency piezoelectric unimorph and bimorph are also discribed as the devices using giant k31 single crystals.
2. Giant electromechanical coupling factor of k31 mode and piezoelectric d31 constant in Pb[(Zn1/3Nb2/3)0.91Ti0.09]O3 single-crystal plates Ferroelectric single crystals made of compounds such as Pb[(Zn1/3Nb2/3)0.91Ti0.09]O3 (PZNT91/09) have been attracting considerable attention, because of the large longitudinal-
4
Ferroelectrics - Applications
mode electromechanical coupling factor of k33 over 92%. Since high-quality and large crystals are necessary to develop devices such as transducers for medical use, we have undertaken and succeeded in the fabrication of PZNT91/09 single crystals with large dimensions. In addition, for further applications to sensors and actuators, a large k31 (d31) mode as well as a large k33 (d33) mode are needed. 2.1 Single-crystal sample preparation The single crystals evaluated were grown by a solution Bridgman method with a Pt crucible supported at the bottom by a conical insulator stand. The crystals without Pt contamination from the crucible have the dimensions of 50 mm (2 inches) diameter, 35 mm height, and 325 g weight. The as-grown single crystals were cut along [100] of the original cubic direction confirmed by X-ray diffraction and from Laue photographs. The single-crystal samples with dimensions of 4.0W(width)x13L(length)x0.36T(thickness) mm for k31, kt and d31 and 4.2Wx4.2Lx12T mm for k33 and d33 were prepared to evaluate the dielectric and piezoelectric properties. Gold electrodes for the following DC poling and electrical measurements were fabricated by conventional sputtering. Poling was conducted at 40 ºC for 10 min by applying 1.0 kV/mm to obtain resonators with various vibration modes. 2.1.1 What is “giant k31 piezoelectricity”? Figure 1 shows the frequency responses of the impedance in k33 and k31 modes in the cases of various coupling factors. It is easy to explain the wide frequency band, which corresponds to the difference between anti-resonant frequency (fa) and resonant frequency (fr), in higher coupling factors. An early work on a PZNT91/09 single crystal poled
Fig. 1. Frequency and phase responses of (a) k33 and (b) k31 fundamental modes in the cases of various coupling factors.
Giant k31 Relaxor Single-Crystal Plate and Their Applications
5
along [001] of the original cubic direction found that the values of k31 (d31) and k33 (d33) modes were 62% (-493 pC/N) and 92% (1570 pC/N), respectively. In a more recent work, the k31 (d31) mode of 53% (-1100 pC/N) and the k33 (d33) mode of 94% (2300 pC/N) were reported for Pb[(Zn1/3Nb2/3)0.92Ti0.08]O3 (PZNT92/08) single crystals poled along [001]. There are significant differences in the k31 and d31 modes between our result and the previous results, despite finding almost the same k33 (95%) and d33 (2500pC/N). A large difference between k31=80.8% and -d31=1700 pC/N in this study and k31=53-62% and -d31=493-1100 pC/N in the previous studies for the single crystals is considered to be due to the following. It is well known that dielectric and piezoelectric properties are strongly affected by the quality of the crystals. It was pointed out that a small portion of opaque parts in the crystal wafer significantly reduces the electromechanical coupling factor of the crystals. Since PZNT91/09 crystals evaluated in this study have very high transparency with a minimum defect level thus far reported, due to better control of the Bridgman crystal growth, the highest k31 and d31 can be obtained. 2.1.2 Where does “giant k31 piezoelectricity” come from? Figures 2(a) and 2(b) show the temperature dependences of k31, k33 and elastic compliance (s11E). Higher k31 and k33 were obtained in the rhombohedral phase below 80 ºC. The values of s11E in the rhombohedral phase are larger than those in the tetragonal phase. Furthermore, the frequency constant (fc=frxL, where L is length), which corresponds to half the bulk wave velocity, of the k31 mode (fc31 =522 Hz·m) is relatively small in comparison with that of lead zirconate titanate (PZT) ceramics (fc31=1676 Hz·m). We believe that the high piezoelectricity in the PZNT91/09 single crystal is due to the mechanical softness of the rhombohedral phase, not the existence of a MPB, for easy deformation by the poling field. This concept may be supported by the result that high k33 (>90%) independent of the rhombohedral composition, such as PZNT91/09, PZNT92/08 or Pb[(Zn1/3Nb2/3)0.955Ti0.045]O3 (PZNT95.5/4.5), was obtained.
Fig. 2. Temperature dependences of (a) electromechanical coupling factors (k31 and k33) and (b) elastic compliance (sE11) in PZNT91/09 single crystal.
6
Ferroelectrics - Applications
In conclusion of this part, frequency spectrum analysis of the responses to vibration modes was carried out in detail utilizing large PZNT91/09 single crystals of high quality. Giant k31 and d31, as well as k33 and d33, were obtained by efficient and uniform DC poling, it means a mono-domain structure in the single-crystal plate as described the next part. The crystals with these giant k31 and d31 will be applied for use in sensors and actuators with high performance. 2.2 Origin of giant piezoelectricity in PZNT91/09 single-crystal plates In order to clarify the origin of giant piezoelectricity in PZNT91/09 single-crystal plates, the poling field dependence of dielectric and ferroelectric properties were investigated in singlecrystal samples with demension of 4.0Wx13Lx0.36T mm for k31, kt and d31 and 4.2Wx4.2Lx12T mm for k33 and d33. The poling field dependence was carried out as follows; the poling was conducted at 40 ºC for 10 min while varying the poling field (E) from 0 to 2000 V/mm. After each poling, the dielectric and piezoelectric properties were measured at room temperature using an LCR meter, an impedance analyzer and a d33 meter. Moreover, the domain structures were observed under polarized light with crossed nicols by an optical microscope. 2.2.1 DC poling field dependence of dielectric constant, d33 constant, k31, kt, k33 and their frequency constants of fc31 and fct Figure 3 shows the effect of DC poling field (E) on dielectric constant (ε33T/ε0) and piezoelectric d31 and d33 constants. There were three stages in ε33T/ε0 with increasing E. The first stage was E>C(t), connected in series with the sample 1. Then the signal was amplified (2) and supplied to the analyzer 3 (e.g., oscilloscope). Here t
φ (t ) = Q(t ) / C m , where Q(t ) = ∫ I (t )dt 0
is the total charge, I(t) – total current, t<tp, where tp is the voltage pulse duration.
(12)
47
MEMS Based on Thin Ferroelectric Layers
Fig. 13. A schematic representation for the method of capacitance and conductance current separation. 1 – a sample, consisted (schematically) of time-dependent values : C(t) and R(t). C0 is the measuring capacitance, φ(t) is the measured potential; 2 – amplifier; 3 – oscilloscope; 4 – voltage pulse generator. Since after the end of the voltage pulse the time Δt necessary to discharge the capacitor C(t) is short and does not depend on the mechanism of the discharge, we have tp
φ1 (t p ) = φ (t p + Δt ) =
∫ Ic (t )dt 0
Cm
= Qc (t p ) / C m ,
(13)
where Ic and Qc are the conductivity current and it’s integral, respectively. Then
( ) ( ( )
( ))
Q cap t p = ϕ t p − ϕ1 t p C m and
С ( t ) = Q cap ( t ) / V ,
(14)
where Qcap is the charge accumulated at the capacitance C(t),V is amplitude of voltage pulse. Thus, by measuring Q(tp) and Qc(tp), it is easy to determine C(t), Ic(t), Icap(t), where Icap(t) is the capacitance charging current. Thus, we separate the conductivity current and capacitance charging current. The macromotion of the slider was measured using the optical microscope, and the time during which this motion occurred was measured by the number of the voltage pulses and their frequency.
7. Experimental studies of thin-film petal micromotors The studies conducted with the samples described above, see (Baginsky & Kostsov, 2007), showed that mechanical and electric characteristics qualitatively correspond to the theoretical estimates for the relatively slightly bent petals. In this case the energy conversion efficiency is 60-70% with the rolling time of 1.5-1.7 ms. But relatively low operating frequency (100 Hz and less, see fig.14, curve 3) causes the micromotor to operate with low power. The resonance frequency at which the slider speed reaches maximum corresponds to the resonance frequency fr of the oscillations of a cantilever with the length l (where l is the petal length):
48
Ferroelectrics - Applications
f r = 0.162 dp / l 2 (EY / ρ )1/2 ,
(17)
where dp is the thickness of the petal, EY is the Young modulus, ρ is specific weight. With the petal thickness of about 1.5-2 μm and the Young modulus for berillium bronze EY=1011 N/m2, fr = 40-60 Hz.
Fig. 14. The frequency dependence of the sample velocity for the cases of bent petals in cross- sectional view (1, 2) and straight petals (3). V=50 V: (1), (3) and V=30 V: (2)
a.
b.
Fig. 15. A view of the bent petal at the position of it's contact with the lower substrate (a) and a scheme for the contact surface between the petal and lower substrate (b). 1 - petal, 2 – the surface of lower substrate (3). Thus, the most important problem is to increase the clock frequency of the micromotor operation. One of the possible solutions is to use the 3D petals structure, when the radius of curvature of the petal cross-section is comparable to petal width b. The cross-section of this petal (1) at the point of contact with the lower substrate (2) is shown on fig.15a. In this crosssection, the petal acts as a spring with the curvature radius r>b. The lateral cross-section of the petal is straight, with the exception of the contact area with the stator, where the petal is bent (see fig. 5F). During the electrostatic rolling of the petal, this spring is attracted to the ferroelectric surface, and its resonance frequency is determined by its width b, and, if it’s length l>>b, is almost independent on length and the applied voltage. It was checked experimentally – the resonance frequency is close to the fr value obtained when l in equation (17) is replaced with b, see fig. 14, curves 1, 2, in contrast to the straight petals, when the fr value is determined by their length (curve 3).
49
MEMS Based on Thin Ferroelectric Layers
a.
b.
Fig. 16. Power (solid lines) and energy conversion efficiency (dashed lines) as a function of load for the U-shaped (a) and flat (b) petals with the mass loading (curves (1)) and the friction loading (curves (2)). The number of petals N=40, f= 1 kHz (a), f=100 Hz (b). Moreover the experiments revealed an unusually fast separation of the metallic films (petals) from the ferroelectric surface after the end of the voltage pulse action, discussed above in Sect. 2. Thus, two factors are identified that can increase the operating frequency of the micromotor. The first one is fast separation of the surfaces; the second one is connected with the 3D petal structure. The load parameters of the micromotor with 3D petals for the operating frequencies that are close to the optimal ones are shown on figs. 16a and 17. With the mass load, one or more clearly identifiable power peaks were observed, that can be explained by inertia properties (fig.16a, curve 1 and fig.17a, b). The comparison of the load properties of the micromotor for the mass and friction loading, see fig. 16a, have shown that the application of the friction loading should increase power and efficiency (η) of the electromechanical energy conversion. With the friction load, the power was independent on the load value at high enough loads (fig.16a, curve 2). Under these conditions with sufficiently large load, the motor abruptly stopped with further load increase. The power peak (mass load) or plateau (friction load) correspond to switch from the inertial mode to the step mode. In the step mode, the motor comes to stop between the voltage pulses.
a.
b.
Fig. 17. The multiple peak structure of load characteristics at mass loading.
50
Ferroelectrics - Applications
The relatively low efficiency in the case of U-shaped petal is explained by the fact that in the resonance determined by the petal width only a small part of the petal length comparable to its width participates in the rolling in the stationary periodic mode, and the rest of the petal length spreads along the ferroelectric surface and is periodically attracted to and repelled from it without participating in energy conversion process. So the petal takes the shape of “bulldozer knife”. Higher efficiency can be achieved with the flat petals, but the voltage pulse repetition frequency consistent with the resonance frequency becomes lower, relative pulse duration decreases, and specific power is reduced despite increase in the specific energy, see fig. 16b. Thus to achieve maximum mechanic power and electromechanical energy conversion efficiency the rolling time must be consistent with pulse repetition frequency. This can be achieved by selection of optimal size and shape of the petals. It can be concluded from the reasoning above that to achieve maximum efficiency the petals must be flat, and to achieve maximum power their length should be decreased. In particular, for operating frequencies on the order of 1 kHz their length should be about 1 mm. The examples of the slider acceleration as a series of voltage pulses is applied are shown on fig.18 for inertial (a) and step mode (b, c). Here, the pulse length is 0.4 ms, and the pulse repetition rate is 1 ms. In both cases, the slider speed would come to plateau at the second pulse at high energy output (with high V). As V decreases, more pulses are necessary for the complete acceleration, see fig.18a.
Fig. 18. Slider acceleration in the inertial mode (a) and in the step motion (b). N = 40. (a): M=0.5g, V=40 V, (b): M=5g, V=50 V and (c) time variation of the sample capacitance at step mode regime corresponding to Fig.18b (first two steps).
MEMS Based on Thin Ferroelectric Layers
51
Finally, let’s discuss the method of increasing the energy conversion efficiency and power output. Fig. 18b shows the acceleration of the slider in the step mode. The acceleration time depends on the amplitude V. The fact that both power and energy conversion efficiency are higher starting from the second, third, etc. pulse (see fig.18 b, c) means that petals tension increases in the equilibrium step mode. The increase in petal tension can also explain the power increase as the load increases. Thus, after the end of the acceleration stage the efficiency increases and can reach rather high values. In the equilibrium mode significant fraction of the petals capacitance is not used for the energy conversion, as part of each petal’s surface remains parallel to the stator surface because it does not have time to come to the initial state, thus assuming the shape shown on fig. 5c. This shape can be visually observed with the continuous slider motion. On the curves showing C(t) as a function of load this petal shape manifests itself through the large part of the capacitance that is independent on the load value (see fig.17b). Thus, efficiency and power dependences on the load are similar (see fig.16, dotted line corresponds to curve 1P(F) for the mass load) and fig.17. Significantly more rapid C(t) growth starting from the second, third etc. voltage pulses (see fig.18c) can also be explained by the electrostatic attraction of the part (about 50% according to the estimates) of the petal that did not significantly separate from the ferroelectric surface in the pause between the pulses moving almost parallel to this surface. So when the next voltage pulse is coming this part of the petal is attracting to the ferroelectric surface much more rapidly compared to the first pulse. The accumulation of the space charge in the ferroelectric leads to the decrease in the efficiency of the energy conversion. This accumulation is connected with both polarization and the injection of the charge carriers under the action of the voltage pulse. After the end of the voltage pulse the electric field still exists on the ferroelectric surface. It attracts the petal and does not allow the petal to assume the initial state. Thus, it interferes with the slider motion in the pause between the voltage pulses. Besides, when the next voltage pulse comes to the ferroelectric surface, the residual potential on the ferroelectric surface when added to the applied voltage decreases the electric field in the metallic film – ferroelectric surface gap. Both these factors lead to deceleration of the slider, and the decrease in the motor power. One of the ways to eliminate the space charge accumulated during the action of the main voltage pulses is the application of the additional pulses (AP) of the opposite polarity in the pauses between the main pulses. Similar processes to alter the potential of the surface at the semiconductordielectric boundary with AP are used in the MNOS memory elements (Yun, 1974). The configuration of the voltage pulses is shown on fig.19d, and the slider speed as function of the additional pulse amplitude V1 is shown on fig.19a. As the shift between AP and the main pulse grows, its effect on the slider speed increase disappears, see fig.19b. The data shown on fig.19 can be explained by the two phenomena: the compensation of the space charge and the deceleration of the slider due to the application of voltage pulse V1, even if for a short time. When AP is applied right after the main pulse, the deceleration plays positive role, leading to the separation of some part of the petals from BSN surface even with applied AP. The latter effect manifests itself through the decrease in the capacitance C some time t after the start of AP (see fig.19c). The following AP parameters were chosen experimentally: amplitude V1= -17 V, duration t1 = 50 μs. With these AP parameters the speed of the slider and the micromotor power are by a factor of 1.5- 2 higher as compared with the mode when AP is not used. The main reason for the decrease in the energy conversion efficiency as the petal assumes the form shown on fig.5C, which excludes part of the petal from the energy conversion process is the mismatch between the frequency corresponding to the motor’s maximum power and the natural frequency of the petal’s vibrations. For example, for the petals
52
Ferroelectrics - Applications
Fig. 19. The characteristics of the motor behavior under conditions, when additional voltage pulse (AP) is applied. a – the dependence of slider velocity on the amplitude of AP – V1 (configuration d), b - the dependence of slider velocity on the delay between main and additional pulses – τd (configuration e), c – the dependence of the structure capacitance on the AP duration. shape shown on fig. 15a with the operation frequency of 1 kHz, the on-line time ratio of the pulses was 0.5, yet the efficiency was less than 20-25%, because the resonance of that frequency corresponded to petal’s width. In case of flat petals, the efficiency was 70-80%, but the operating frequency decreased to 100-40 Hz, which resulted in the on-line time ratio increase by a factor of 10-20. Thus, despite the increase in the mechanical energy generated during one conversion cycle by a factor of 3-4, the power was fallen down by a factor of 3-5 The second reason for the incomplete conversion of the rolling capacitance energy into the mechanical energy is the incomplete rolling of the petal’s surface, that is, the formation of the so called rolling needle shown on fig.15b. The formation of this needle can be attributed to the difference in the speed vl of the longitudinal (along the petal’s length) rolling and speed vb of the lateral rolling, because initially the petal has the 3D structure shown on fig.15a. This results in the motive force decrease, because in this case only the part of the petal’s width that is on the front end of the rolling contributes to the driving force (see fig.15b). It can be noted that this phenomenon can explain the possibility of the significant efficiency increase with the practically unchanged power by increase in the petal’s stiffness achieved by small increase in the thickness. In this case the efficiency increase with the rolling capacitance decrease can be attributed to the decrease in the unused capacitance. This happens because relatively smaller part of the lateral surface of the petal is rolled because of the increase in the lateral stiffness. The above mechanism can explain steep
MEMS Based on Thin Ferroelectric Layers
53
(faster than quadratic) growth of P as a function of V at the low voltages. At high voltages, P is quadratic function of V, because the growth in voltage increases the lateral rolling speed and leads to the rolling needle disappearance. The decrease in the petal length accompanied by appropriate (roughly proportional) decrease in the gap thickness de allows one to achieve the resonance motion of the slider with respect to the petal length, and thus achieve high efficiency accompanied by the specific power increase. To increase power by an order of magnitude it is necessary to use 0.5 mm and shorter petals.
8. The peculiarities of ferroelectric ceramic application in petal micromotors To create micromotor prototype thin ceramic plates were used made of PZT material with the composition Pb0 - 66%, ZrO2 - 21%, TiO2 - 11% and dielectric constant of εF ~ 3900. Also, we used plates made of antiferroelectric ceramics with the composition close to PZT and dielectric constant of 10000. The surfaces of the ceramic plates used for the electrostatic rolling were polished up to the optical smoothness (roughness of about 10-8 m). The metallic electrode (silver film with 1 μm thickness) was applied to another ceramic surface by vacuum deposition followed by sintering. One of the peculiarities arising from the use of the ferroelectric ceramics in the described construction, as opposed to the use of barium-strontium niobate films (Dyatlov et al., 2000; Baginsky & Kostsov, 2003) is higher value of the switched polarization part and longer time before polarization disappearance. To significantly decrease the effect of the polarization switch in the step mode, it is necessary to apply the pulse of the opposite polarity with length and amplitude sufficient to bring the polarization direction into its initial state before the application of the slider moving pulse. But even this scheme of voltage pulse application does not completely solve the problem of polarization screening charge, since between the pulses there is an electric field near the ferroelectric surface that causes the slider to stop and decreases the motor power. The investigations of micromotors made on the basis of PZT ceramics revealed only a small values of mechanical energy and specific power because of high values of polarization damping the motion. The effect of the polarization processes can be significantly decreased by using antiferroelectric materials with high ε, that are shown to have quite small or no residual polarization (Burfoot & Taylor, 1979). The ceramics was 100 μm thick, with the specific capacitance of about 900 pF/mm2. The specific capacitance during the electrostatic rolling was only 20 – 40 % smaller than that of the ferroelectric films despite higher thickness. Since the use of AFE ceramics allows one to apply significantly stronger voltage pulses than would be possible for the ferroelectric films without compromising the operation reliability, it is obvious that the energy capacitance and motor power can also be significantly increased. Fig. 20 shows the frequency (a) and load (b) properties of the micromotors based on the AFE ceramics for the constant number (n=40) and size (3.5*0.5mm) of petals and their dependence on the voltage pulse duration (c). To eliminate the effect of the space charge that is created by the leakage current caused by the voltage pulse, after the end of the main pulse (MP) the additional pulse (AP) was applied with the amplitude equal to that of the MP, but of the opposite polarity and with smaller duration (100 μs). This relationship between MP and AP parameters was found experimentally to maximize the power of the micromotor with the given load.
54
Ferroelectrics - Applications
The analysis of the frequency properties of the motor power (fig. 20a) showed the resonance at clock frequency of about 1 kHz. The resonance position was virtually independent on voltage pulse amplitude and motor power and load. It shows that the peak can be mainly attributed to the mechanical resonance based on the petal width. For the fixed load and voltage pulse amplitude the micromotor power can be maximized by adjusting the pulse duration and the time between the pulses. For example, fig.20c shows the typical curve of micromotor power as a function of pulse duration, with the optimal pulse duration of 450 μs. The complex shape of this curve with two maximums is explained by manifestation of two effects. The first peak appears due to the mechanical resonance of the petal at frequency of about 1 kHz (Baginsky & Kostsov, 2003). As the pulse duration tp grows at the conditions of fixed gap between the pulses Δt the additional part of the petal’s length is involved in the process. It gives rise to the additional grows of power despite of the frequency f = 1/T (where T = tp + Δt) decrease, so the second peak is forming. Load curves, fig. 20b show the power peak at 40 g load (the friction coefficient k is 0.3). The maximal power was 1.5 mW, which is 2-3 times greater than for the similar motor based on the ferroelectric films.
b.
a.
c. Fig. 20. Power and speed of the antiferroelectric ceramics motor, (а) – as a function of voltage pulse period (m = 40 g, tp = 0.45 ms), (b) – of the load mass (T = 1 ms, tp = 0.4 ms) and (c) - of the voltage pulse duration. m = 40 g, V = 85 V, t1 = 0.1 ms. Table 1 compares the maximal absolute and specific power, P and P1, with the same mass load, friction coefficient (k=0.3) and number of petals (n=40) for motors based on barium-strontium niobate (BSN) films (Ba0.5Sr0.5Nb2O5 composition), PZT-ceramics and AFE-ceramics.
55
MEMS Based on Thin Ferroelectric Layers
Material BSN films PZT-ceramics AFE- ceramics
P, μW 500 70 1500
Petal size, mm 3.5*0.5 3.5*0.5 3.5*0.5
P1, μW/mm2 7.14 0.83 21.4
V, volts 50 90 85
Table 1. A comparison of power of petal electrostatic micromotors on different materials. Micromotor power can be increased by decreasing d down to 50-20 μm, or by increasing voltage pulse amplitude up to the breakdown voltage in the gap between the petal and the ferroelectric surface. According to estimates in (Dyatlov, et. al., 1996), this voltage can be as high as 200 V. Besides, the power of the linear motor can be increased by increasing the operation clock frequency, the optimal value of which in turn depends on resonant frequency of the petal as have been shown above. Thus the duration of the separation process does not affect the frequency properties of the motor, and clock frequency is limited by the duration of the electrostatic rolling process. The experimental clock frequencies for the ferroelectric films are in 10 – 20 kHz range (Dyatlov et al., 2000). Thus, the maximum specific power of the “ceramic” motor can be estimated to be equal to 100 – 300 μW/mm2.
9. Some applications of the micromotors High energy output allows one to obtain high absolute power (up to 0.01 – 1 W) increasing the rolling area, and therefore the described micromotors can be used in various MEMS devices, e.g., listed in Introduction. Some applications of proposed micromotors in MEMS were analyzed by us both numerically and experimentally. The possibility of creation of high speed (microsecond range) microcommutators powered by microactuator based on an electrostatic rolling of the thin metallic film on the ferroelectric film surface was considered in (Kostsov & Kolesnikov, 2007). The numerical analysis of the microcommutator operation was performed and its main characteristics were described. It was shown that the driving force developed by the microactuator in the first 10–100 μs of the electrostatic rolling is equal to 0.05–0.5 N per 1 mm of metallic film width, and the force is limited by the mechanical strength of the film. The high value of the force makes it possible to use strong springs that prevent the switchboard from switching between steady states even under the load factor of 1000 g and more The research on opportunities of construction of high – efficiency micropumps and injectors of liquid microjets on the base of high energy-intensive electrostatic microactuators, working in a cyclic mode, was carried out (Kostsov & Sokolov, 2010). The design, the features of functioning, characteristic parameters of such devices are described. It is shown that a microactuator with the area of 1 mm2 is capable to inject during one step with the duration of 30-300 μs a microjet of liquid with the weight of 1 - 3 micrograms, flowing out with the velocity of 1-10 m/s and more depending on the radius of exit nozzle. Electrostatic high energy micromotor based on the ferroelectric films is studied as applied to microelectromechanical devices operating in vibrational mode (Baginsky et al., 2008). It is shown that the micromotor can be efficiently used in high frequency micromechanical vibrators that are used in high energy MEMS devices, such as micropumps, microvalves, microinjectors, adaptive microoptic devices etc.
56
Ferroelectrics - Applications
The operation principle of micromechanical valve based on the effect of electrostatic rolling of metallic films on the ferroelectric surface was considered (Kostsov & Kamishlov, 2006). These microvalves differ from prototypes by high operation speed (microseconds), by ability to sustain a high pressures and by good fabricability. Finally, the preliminary experiments and numerical modelling have shown that these microactuators can be used as microelectrogenerators with wide application range by reversing the described electromechanical energy conversion (Baginsky & Kostsov, 2002).
10. Results and discussion The comparison of the operation parameters for the different types of the electrostatic micromotors is shown in table 2. Here AR = CspV2/2 is the electric work during one cycle. Mechanical work is AM = ηAR. The energy conversion efficiency η value is determined by the micromotor construction. For the first two types of the micromotors relatively high efficiency values are achieved – up to 80%, for the micromotor described in this paper the efficiency can also be as high as 80% depending on the petal geometry. The analysis of the data in Table 2 shows the specific mechanical work of the described micromotors to be greater than AM values typical for the known constructions of the micromotors used in MEMS by several orders of magnitude. Micromotor type Air gap Rolling on the linear dielectric Rolling on the ferroelectric
d, μm
C1, pF/mm2
2-3 2-3
1 Tbit/inch2) nonvolatile memory devices (Vettiger et al., 2002; Pantazi et al., 2008; Hamann et al., 2006; Ahn et al., 1997; Cho et al., 2003; Cho et al., 2005; Ahn et al., 2004; Cho et al., 2006; Heck et al., 2010). In such a system, an atomic force microscope (AFM) probe (or an array of AFM probes) is used to write and read data on a nonvolatile medium; the bit size depends mainly on the radius of the probe tip. Moreover, the storage area is not defined by lithography like in SSDs, but rather by the movement of the probes. Thus improving the probe motion control to the tenth of a distance can translate into two orders of magnitude higher density. Bit size as small as 5 nm and a storage density in the Tbit/in2 regime with data rate comparable to flash technology have been achieved (Cho et al., 2005; Cho et al., 2006). Unlike SSD technology which requires new lithographic and fabrication tools for each new generation, manufacturing of the probe-based device can be achieved using existing low-cost semiconductor equipment, which can reduce the price of these devices considerably. Another advantage of probe-based memory is that the mechanism to move the probes is low power, which reduces power consumption and heat dissipation in comparison to HDD devices. While various writing mechanisms have been proposed for probe-based storage, e.g., thermomechanical and thermal writings on polymeric and phase-change media (Vettiger et al., 2002; Pantazi et al., 2008; Hamann et al., 2006), a great deal of attention has recently been devoted to the electrical pulse writing on ferroelectric films due to the non-structuredestructive nature of the write-erase mechanism (Ahn et al., 1997; Cho et al., 2003; Cho et al., 2005; Ahn et al., 2004; Cho et al., 2006; Heck et al., 2010). When a short electrical pulse is applied through a conductive probe on a ferroelectric film, the highly concentrated electric field can invert the polarization of a local film volume, resulting in a nonvolatile ferroelectric domain that is the basis of data recording. This mechanism allows for longer medium lifetime, i.e., larger number of write-erase cycles that is comparable to hard disk drives, faster write and read times (Forrester et al., 2009), smaller bit size (Cho et al. (2006) and higher storage densities (Cho et al. (2006). Although the probe-based storage technology based on ferroelectric media has shown great promise, no commercial product has yet reached the market. This is mainly due to
158
Ferroelectrics - Applications
fundamental limitations of the media material and probe-media contact during probesliding. For ultrahigh storage density exceeding 1 Tbit/inch2, domain size reduction below 10 nm is required. Small domain sizes can be obtained by decreasing the size of the probe tip. However, the inverted domain is subjected to ferroelectric depolarization charges and domain-wall energy (Li et al., 2001; Wang & Woo, 2003; Kim et al., 2003) that can be high enough to invert the domain back to its initial polarization. It has been predicted (Wang & Woo, 2003) that inverted ferroelectric domains smaller than 15 nm are unstable and could be inverted back to their initial state as soon as the electric pulse is removed. This instability can be further exacerbated by the presence of a built-in electric field due to film defects present in thin ferroelectric films, which is anti-parallel to the inverted domain polarization. In short, this fundamental instability has prevented the demonstration of stable inverted domains less than 10 nm in size in ferroelectrics. Reading such sub-10 nm inverted domains at the required high speed and with high signal-to-noise ratio (SNR) is also another important issue as such a technique has to be suitable for a MEMS-based probe storage system (Heck et al., 2010). Another technological bottleneck is that the high data access rate requires a probe-tip sliding velocity on the order of 5 to 10 mm/s, over a lifetime of 5 to 10 years, corresponding to probe-tip sliding distances of 5 to 10 km. The bit size, and thus the storage density, mainly depends on the radius of the probe-tip that is prone to rapid mechanical wear and dulling due to the high-speed contact mode operation of the system (Cho et al., 2006; Knoll et al., 2006; Bhushan et al., 2008; Gotsmann et al., 2008). This tip wear causes serious degradation of the write-read resolution over the device lifetime. In this chapter, we review solutions that have been proposed in the literature to address the above fundamental issues and that will enable the development of probe-based nonvolatile memories with storage densities far exceeding those available in today’s market. This chapter is divided into four parts. In the first part, the relevant theory and mechanism of pulse-based writing as well as probe-based storage technology on ferroelectric media are reviewed. The stability of single-digit nanometer inverted domains is addressed next. Reading schemes at high frequency and speed are then discussed. Finally a wear endurance mechanism, which allows a conductive platinum-iridium (PtIr) coated probe-tip sliding over a ferroelectric film at a 5 mm/s velocity to retain its write-read resolution over a 5 km sliding distance, is reviewed.
2. Background Ferroelectric materials such as BaTiO3 and Pb(Zr0.2Ti0.8)O3 (PZT) have a perovskite crystal structure in which the central atom (Ba/Zr/Ti) is bi-stable and can be shifted up or down by applying an external electric field (Figure 1a) (Ahn et al., 2004). Upon removal of the external field, the new atom polarization remains, resulting in a nonvolatile property, which is the basis of data recording. To shift the polarization of the central atom, a probe tip can be used (Figure 1b). By contacting the probe tip to the ferroelectric film and applying a bias pulse between them, a highly concentrated electric field underneath the tip is created which flips the polarization of a local volume of atoms and form an inverted polarization domain that can be used as bits for data storage (Figure 1c). The bit can be erased by applying a pulse of a reverse polarity which will switch the polarization within the written domain (Figure 1d) (Cho et al., 2003).
Ultrahigh Density Probe-based Storage Using Ferroelectric Thin Films
159
Fig. 1. Data storage on ferroelectric media. (a) Crystal structure of the perovskite ferroelectric PZT showing upward and downward polarization variants. (b) Schematic of bit writing using a probe tip to which a voltage is applied. (c) 4×4 inverted domain dot array formed on a ferroelectric medium. (d) Selective erasing of domain dots by applying a bias of reverse polarity. The size of the volume mainly depends on the sharpness of the probe tip. In principal, the inverted volume can be as small as an individual atom, and thus allowing for a single atom memory (Ahn et al., 2004). Therefore, an ultrahigh density memory can be constructed with such a system if ultra-sharp probe tips are used and cross talk between bits is avoided. In fact, bit sizes as small as 5 nm (Figure 2a) and a storage density of 10 Tbit/in2 with an 8 nm bit spacing have been achieved (Figure 2b) (Cho et al., 2006; Cho et al., 2005). Such a storage density is by far the highest ever achieved in any storage system. Moreover, domain switching times can be as fast as 500 ps, allowing for high writing rate (Figure 2c).
(a)
(b)
(c)
Fig. 2. Nanodomain formed using pulse writing on ferroelectric media. (a) Smallest nanodomain reported in the literature (Cho et al., 2006). (b) Highest writing density ever achieved corresponding to 10 Tbit/in2 (Cho et al., 2006). (c) 500 ps long pulse used to fully invert nanodomains in ferroelectric media (Cho et al., 2006).
160
Ferroelectrics - Applications
Following the IBM Millipede and HP ARS systems, a joint team at Intel and Nanochip (a startup company) has recently developed a device named “seek-and-scan probe (SSP) memory device” in which the pulse writing scheme using ferroelectric media is used (Heck et al., 2010). The device architecture is shown in Figure 3 and consists of three layers. The bottom layer contains an array of 5000 MEMS cantilevers with tips that are directly fabricated on CMOS circuitry. The cantilevers are spaced at a 150 µm pitch, corresponding to the stroke of the electromagnetically actuated x–y micro-mover which forms the second layer of the device with the ferroelectric media film grown on its lower side. The third layer is a cap wafer that seals the device. The device is 15.0×13.7 mm2 in size and consumes less than 750 mW with a maximum of 5% related to the MEMS actuation. It is capable of achieving a data rate of 20 Mbyte/s using 272 read-write channels. This rate is the highest ever reported in probe-based devices. The MEMS cantilevers are fabricated directly on standard Al-backend CMOS in order to increase the overall signal-to-noise ratio (SNR) of the device. This is achieved by growing a low temperature (